I need help.
I am using the MACH3 3,041 in all reversals of direction (in any axis), but only for a sense (+), when the signal from the DIR has to go down, my servo motor account 1 pulse to the wrong side, then again from the right side, losing 2 pulses, which in my case = 0.002 by the pulse = 0.004mm in total.
When I turn the output STEP and DIR to perfect it, but the step can not pass a certain frequency, or at least when greater frequency, (high speed), my driver not to recognize.
I use Servo Drives Panasonic and a Bob C11G of CNC4PC - Duncan.
My driver has inputs with opto.
I already tried the following:
. Increase the time of "STEP" until 20us, not resolved, the system is held in high speeds.
. Increase the time of "DIR" until 20us  not resolved.
. Other versions of MACH3 until the "View 2.00", not resolved.
Please make any ideas for me?
I think the time reversal of direction is not large enough to make my driver a reversal. But a sense works, how it happens?
Is there any parameters possible to adjust the difference between Step and DIR?
What is the difference between the first "DIR" pulse and the first "STEP" You can adjust this parameter?
Thank you.

I don't really understand what your asking. Do you have backlash enabled? If so, disable it and see if it helps.

Brett

I think i understand that you are loosing position when reversing direction at a high speed , please correct if we are wrong etc

this could be a number of problems and may well be difficult to trace , although i would guess depending on your servo drives that they are matched , that it may be the opto's not switching fast enough , it may be you need better opto's fitted , this would need investigation further with a scope , along with motor tuning .

you are also getting into the area of high tolerance parts such as your ballscrews and bearings , are they better than 0.0004 ?
and can they take the stresses of changing direction at speed ,both radial and axial directions , and are they preloaded correctly .

work your way through everything mechanicaly first , as any inaccuracys will add up along the way
 

Thank you,

I have used low accelerations and nothing resolved.
With the scope I saw that the pulse of Step arrives simultaneously or very close to the pulse of DIR.
And the mechanics are OK. Through the software I can find out the quantity of pulses that arrive in the driver and compared with oscilloscope, they are perfect, in fact they come in correctly, but its direction is wrong.
So I think there should be a delay time between  the pulse of DIR for the pulses of STEP, because my case is that in one direction (1 - 0) the first pulse STEP still serves for the previous direction.

Thanks,

Prendice

Try wiring your step and dir wires DIRECTLY to the driver board, and miss out the BOB board.

We have had previous posts where CNC4PC have had funny results. I think it is the Opto electronics that slow down the pulse.

Try a GOOD quality parallel cable, it needs to have one actual ground (common) wire per I/O pin. Trying to send a bunch of digital signals down a length of tightly (physically) coupled wires is tricky. The pulses will will be slewed at the receiving end. A high-quality calble as short as possible is a must.

Thank you,

I will try to work without the BOB and see the result today.
I will inform them the result.
I hope solve my problem, but after I talk with Duncan.
Maybe he can solve the problem with new components.

Thanks,

Prendice

The signals without BOB C11 were perfect and do not have the problem with the loss of 1 pulse of STEP.
I have 2 more C11 V.2.0, I will make the test with these
The C11 V. 3.1 did not work.
I explain the case to Duncan.
Again thanks for your help.
I'm still starting with the MACH3.
My new experiment will be with Smooth S.
I hope good results, if possible the help you also.

Thank you.

Prendice

You will find that an UNPOWERED BOB board will be fine. There is no interference with the signals, and they are cheaper.

Provided you have a SEPERATE 5 volt power supply for your home and limit switches, and your M3,M4 M5 relays and coolant, you will be fine. (SEPERATE  from the MOTOR supply, that is).

I would look at the DIR/STEP timing as it comes out of the PC, and also as it goes into the servo drivers.  It's possible the CNC4PC board is changing the timing.  I do know some of the older CNC4PC boards (At least the C11, not sure about the C11G) used very slow optos for the DIR signals, and fast ones for STEP, which could cause exactly what you're seeing.  In my case, the DIR transitions were delayed 30-60 uSec.  I was able to correct this by changing the phototransitor bias resistors on the board, from 10K to 390 ohms.  The newer CNC4PC boards use the fast optos for all signals.  I suggest you contact Arturo at CNC4PC, and let him know about your problem.  He's very good about customer support.  In my case, he immediately shipped me, at no charge, a brand-new, latest rev board to replace mine, even though it was more than two years old.  The new board works perfectly.

Regards,
Ray L.